STAUB, NORMAN C., HITOSHI NAGANO, AND MORTON LEE Reviewers of the extensive literature (2, 13-15, 41)
PEARCE. Pulmonary edema in dogs. especially the sequence of &id have not devoted much attention to these questions.
accumulation in the lungs. J . Appl. Physiol. 2 2 (1) : 2 2 7-240. Visscher, Haddy, and Stephens (41) proposed a theoreti-
1967. -We induced acute pulmonary edema in anesthetized cal sequence : intracellular, “interstitial,” and alveolar
dogs by transfusion of blood and dextran (g dogs) and by edema.* They could not find any data on the intracellu-
intravenous injection of alloxan (I 2 dogs). We measured
lar phase but thought it was a very minor factor. They
several indices of cardiopulmonary function, including pul-
suggested that during the interstitial phase there were
monary capillary blood volume (Vc). Cardiac output, central
blood volume, and Vc decreased; average pulmonary artery changes in the elasticity of the lung and changes in ex-
pressure @pa) increased transiently and left ventricular end travascular pressure influencing the lung vessels. They
diastolic pressure (Plved) remained constant after alloxan. believed, without specifying how, that the occurrence
Output, central volume, Vc, ppa, and Plved increased after of interstitial edema may be important to understanding
transfusions. The data clearly indicate different primary the processes involved. The most serious phase was air-
mechanisms for the edema caused ’ by the two procedures. space filling which blocks aeration of alveoli. They did
Based on studies of the rapidly frozen lungs of these animals, not discuss the temporal interrelations among the various
both types of edema show the same sequence of fluid accumu-
phases.
lation in various compartments of the lung. Fluid appears first
Drinker (I o) investigated the dynamics of pulmonary
in the interstitial connective tissue compartment around the
edema extensively. He did not believe there was a com-
large blood vessels and airways. Alveolar wall thickening fol-
lows. Alveolar filling begins after the interstitial compartment plete alveolar epithelium; therefore fluid leaking from
is well filled. Alveolar filling occurs independently and rapidly the capillaries filled the wall and alveolar spaces more or
in individual alveoli. Air is not trapped. Atelectasis is uncom- less simultaneously even before the lymphatic vessels
mon although the total volume of each fluid-filled alveolus is were filled. He interpreted the widely expanded peri-
reduced. vascular spaces seen in severe edema as grossly distended
lymphatics rather than true tissue spaces. He concluded
alloxan ; capillary permeability; pulmonary alveoli; pul- that diffusion barrier was the main early physiologic ab-
monary circulation; respiration; surface tension
normality but that later airway obstruction was pre-
dominant.
Macklin (22) examined normal movement of fluid
in the lung especially in relation to the clearance of par-
W HAT IS THE SEQUENCE of events by which fluid ac- ticulate matter. He believed there was a complete
cumulates in the lung in pulmonary edema? Is the epithelium that was rather impermeable to water but
sequence independent of the method used to produce the that a protein-free fluid containing a surfactant was
excess of leakage over drainage? Can we gain any insight
into the normal movement of fluid from this abnormal 4 We have placed quotes around this use of “interstitial” be-
situation? cause the reference does not clearly define the term; however, on
p. 393 the authors state, “. . . the only early evidence of lung
Received for publication 23 March 1966. edema that is visible microscopically is interstitial wall thickening,
r This study was supported in part by Public Health Service because protein-free fluid in alveoli is undetectable in conventional
Grant HE-06285 and Los Angeles County Heart Association histologic preparations.” If the authors mean alveolar wall
Grant 322. interstitium, then their usage is different from ours. There are two
2 Senior Fellow, San Francisco Heart Association. Present interstitial compartments in the lung: that between the epithelium
address: Research Institute for Diseases of the Chest, Faculty of and endothelium of the alveolar walls as just mentioned and that
Medicine, Kyushu University, Fukuoka, Japan. around the conducting airways and vessels. We follow the usage of
3 On leave from Dept. of Medicine, University of California von Hayek (I 8) and Ruszynak, Foldi, and Szabo (3 I) in calling
Medical Center, Los Angeles, and Veterans Administration Center, the latter interstitial. Functionally, we regard the alveolar wall as a
Los Angeles, Calif. unit in so far as the diffusion path from gas to blood is concerned.
227
228 STAUB, NAGANO, AND PEARCE
TABLE I. Summary of pulmonary and vascular data for jiue control animals
Esp
I
Period DL Dm vc N-$$ CL Qpe vv Plved Ppa Anatomy*
X0. 2
IC 18.4 Base line+ 11.6 rt 0.4 43 d= I( 24 * 3 152 A= 39 2.0 & 0.2 -2 XII 0.: 7 & 0.E
I 16.9 co 24 57 2.0 -3 7
2 II .6 25 34 83 2.2 -2 6
3 ‘5.7 “4 17 ‘I3 / 2.6 -2 6 No edema
I
2c 19.8 Base line 14.0 zt 0.2 29 * 4 74 + IO 2.4 =t 0.1 -5 + 0s 8 zt 0.3
I ‘4.9 3’ 82 2.4 I -8 9
2 16.9 33 80 2.2 I -7 9
3 13.2 24 74 2.4 j -8 9 No edema
3c 24.2 Base line 9.2 zt 0.3 20 * 3 83 =I= 35 2.6 =t 0.2, 1.73 457 89 7 II= o-5
I 7.3 21 2.1 ’ IO
2 8.2 ‘9 97 2.4 7
3 10.0 16 86 I .40 349 66 8 No edema
4c Base lint 4.8 =t 0.2 32 8 zt 0.2 12 2.0 It 0 I .40 31’ 90 t5 * 0*3 74=0
I 4.8 00 8 42 2.0 1.14 ‘79 69 +5 7
2 5.0 00 7 51 I .8 AZ1 9
3 5.0 co 6 38 I .8 I .22 349 52 *3 IO No edema
Units: \Yt (kg), D+ and Dm (ml/min X mmHg), Vc (ml, corrected to 15 g Hb/Ioo ml blood), A@ - a)DOe (mm Hg), CL
(ml/cm T-T&> X kg), 0, (liters/min), Qi (ml), Qp evw (ml), Plved and Ppa (mm Hg). * Based on gross and subgross anatomy
and clinical findings of tracheal foarn, AQpevw and A(A - a)DOs. t Base line: Data show average and one standard deviation
for three determinations over 45-min period. I, 2, 3 : individual data for three Io-min periods after completion of base line period.
1 11dditional controls of vascular flow and volume data, done after the main series was completed. 8 Mean left atria1 pressure Pla,
actively secreted onto the alveolar surface by the granu- uniformly. Their evidence led them to state that changes
lar pneumonocytes. He believed most of the alveolar in alveolar surface tension were probably not involved in
fluid was exhaled but some flowed to regional lymphatic the genesis of edema by increased capillary hydrostatic
su1nps. How the surface fluid and any alveolar wall pressure.
tissue fluid reached the lymphatics is unclear. He vari- According to Visscher and his associates one reason
ously invoked flow over the alveolar surface, focal re- that so little is known about the bulk movement of fluid
absorption, microcurrents within alveoli, and phagocyto- within the various compartments of the lung, especially
sis. in the earliest stages of edema, is the lack of good methods
Williams (44), Forster (12), and Said and co-workers for seeing the edema process with precision. We have in-
(33) adduced evidence that alveolar filling was the only duced acute pulmonary edema in anesthetized dogs in
physiologically important stage as far as gas exchange is two ways: r) elevation of pulmonary capillary hydrostatic
concerned. On the other hand, Schulz (34), using clec- pressure by rapid infusion of blood and dextran and 2)
tron microscopy, studied the alveolar wall in a variety of rapid intravenous injection of alloxan. We followed the
acute edemas, and concluded that there is a definite early functional changes that occurred and then examined the
stage of alveolar wall swelling due both to intra- and histologic pattern of fluid filling by using rapid freezing
intercellular edema. of the inflated living lung (38), a procedure that fixes
Pattle (25, 26), Clements (8), and Greene (15) have the lung very quickly at a given point in time, thus per-
added the alveolar interfacial surface tension between mitting us to “see” the lung as it was in life.
gas and tissue to the balance of forces that must be con- Our report deals first with the problem of mechanism
sidered in alveolar fluid filling. in the two methods used to produce edema, then presents
Recently, Cook et al. (9) proposed that air-space filling evidence about the sequence of events as fluid fills the
occurred independently in individual alveoli rather than lung.
ACUTE PULMONARY EDEMA 119
TABLE 2. Summary of pulmonary and vascular data for nine dogs with high-pressure edema
- -
DL vc A(koa) 0 Qi Qwvw Plved Ppa
2
-.
Base line 13.9 zt 1.c j3 Zk IC 27 * I .oo rf: 6 3.2 * 0.4 c I .72 392 47 o*o g zt 0.6
I 10.8 25 52 220 2.0 I .25 483 69 33 43
2 10.2 69 34 267 I .8 30 40
3 9.2 56 40 132 I.9 I .2c 474 118 28 37 Moderate ede
2P 20.0 Base line 23.1 rt 0.4 72 * 7 45 I!= 2 64 =k 16 3.6 A 0.2 ). I .8~ 466 63 ozko 9*0
I 23.2 44 70 51 2.6 .13.8: 782 78 46 58
2 20.4 110 47 121 2.5 48 54
3 12.8 30 46 236 2.3 1G9i 787 216 40 46 Severe edema
22.3 Base line 22.0 rt I.9 25 * 2 IO 7.1 * 1.c 1 I .7E 274 78 -1 zt I.2 5 zt 0.6
I 21 .o 41 54 4.3
t2.6~ 552 56 33 26
2 17.2 51 123 4.6 35 28
3 14.4 43 165 4.5 1$.25 612 127 25 20 Mild edema
18.E Base line 13.2 It 0.5 79 * 41 29 * 2 34 + 3 3.6 h o ,2.0: 362 72 3 xi= 0.6 7 zt 0.6
I 7.8 18 42 103 2.0 .I5.5E 544 86 30 25
2 6.5 IO 7’ 85 I .8 28 30
3 5*8 7 ‘45 48 I.5 .I5.64 547 88 22 25 N o gross edema
‘9.4 Base line 11.0 & 0.2 $8 * 5 24 rt I 68 =I= 43 3.1 * 0.4 , -3 zt 0.6
I 9.5 21 32 180 2.1 30
2 9-o 20 32 229 2.1 20
I 6.4 29 14 92 I.9
2 6.2 23 I4 58 I-9
3 5.8 26 14 18 I l9 Xv0 edema
TABLE 3. Summary of pulmonary and vascular data for twelve dogs with alloxan edema
A(A - a)
Wt Period DL Dm vc
DO2 CL 0 Qi Qpevw Plved Ppa Ana tomy
-
‘5.4 Base line 8.9 zt 0.2 22 & II '9 =t 2 160 ziz 44 4.3 Ii= 0 +I zk I* 6 rt I.( 1
I 7.6 16 ‘9 248 3.8 0 14
2 7.7 15 21 395 3.2 0 8
3 7.5 37 12 330 3.1 0 6 Mild edema
20.2 Base line lg.1 * 0.1 38 =J= 3 29 * I 114 =t 18 2.8 =t 0.1 -4 * o* 7 zt 1.c 1
I 9.9 ‘9 25 370 2.0 -4 30
2 5.1 6 33 572 I .6 -4 12
2o.c Base line II.0 * 0.5 x 13 zt I 63 =i= 40 2.1 * 0.2 II * 1.c 8 sfs 0.: i
I 9.7 a-2 7 172 I .8 12 9
2 8.7 w IO 229 I .6 12 9
3 7.5 a0 3 524 I .6 IO IO Moderate
edema
20.6 Base line rg.2 rt 2.c 60 * 21 Fl-4=t= 42 * ‘9 4.0 zt 0.2 2.66 6 A 1.c 1
585 65
I 9.6 23 20 99 3.1 12
2 7.5 16 16 498 2.6 IO
‘9.3 Base line :3.1 * 0.8 17 * 4 30 * 4 25 * 7 3.1 * 0.3 2.14 331 72 o=o
I II .6 31 26 73 3.4 2.30 195 82 9
2 II .2 6 17 65 3.0 5
3 8.7 w 9 92 2.4 I .g2 231 Moderate
‘09 5
edema
16.5 Base line 4.8 A 0.7 p zt 8 32 * 3 34 =I= 15 3.0 Ik 0.1 I .64 554
I II .7 33 23 50 2.4 2.11 381 Rapid, severe
edema
18.3 Base line 12.7 * 1.5 43 * 21 35 A= 5 53 =I= 10 3.1 * 0.2 I .80 304 62 5 zk 1.c 1
I II .I 26 33 163 2.6 I .70 244 62 5
2 10.4 44 24 137 2.4 4
3 9.8 69 21 138 2.2 I .25 176 41 4 No edema
- -
Abbreviations, units, and explanations as in Table I. * Mean left atria1 pressure @la).
the aorta.5 We measured end-tidal CO2 via a catheter in the tracheotomy tube through an infrared CO2 analyzer
(Beckman LB-I). The outputs of the pressure gauges and
5 We did not accurately adjust the reference levels of our
pressure gauges; therefore, the pressures recorded should be used CO2 meter were amplified and recorded on a polygraph
only to measure changes in pressure, not absolute values. (Grass model 5). We gave intravenous injections and
ACUTE PULMONARY EDEMA 231
TABLE 4. Alveolar wall thickness in control and Table 3 (alloxan edema). The three base-line period
tlarious grades of edema values are listed as the average + one standard deviation.
__ _.____~_
-- -~ ~.__--.--- The experimental period data are listed individually. Not
shown are the values for hemoglobin concentration which
Group and Exp So. did not change significantly in the controls, decreased in
Upper zone Lower zone
~~
the high-pressure series as plasma volume was expanded
L_-----
more than red cell volume, and increased in the alloxan
Control
1Uone 5.kk2.0 series as edema developed.
3c
4C None 4.31t1.3 Most of the data are similar to those reported by
others. Cardiac output (Q) and central blood volume
High pressure (Qi) usually increased with the fluid infusions (P series)
8P None 4.o~to.8 6.okr.5
Slight 4.3ztI.3 4.g*1 .o
but fell after alloxan (A series). Average pulmonary
4P
IP Moderate ; 6.1~~3.2 6.8&r .3 artery pressure (ppa) and left ventricular end diastolic
2P Severe ) 5.o~t1.0 7.5~t2.0 pressure (Plved) rose in the high pressure group; f’pa
6P Severe 5.6~11.7 7 -4+1-4 increased slightly in the alloxan group. Lung compliance
did not change in the controls because of the repeated
None 4.0% .2 5.3zt1.6
inflations but did fall significantly in both edema groups.
Mild 3.8&1 .I 5.8&1 .o The A(A-a)DOz d uring high-oxygen breathing serves as
Moderate 6.2ho.g 8.01~ .7 an index of shuntlike effect; it did not change signifi-
Scvcrc 6.~h1.6 8.31tr .4 cantly in the controls but rose with time in both edema
Severe Jj.r*o.g 5*5*0-g
Severe 7.o~t1.5 series.
7*7*0-g
We invite particular attention to the Vc determina-
Values are averages and I SD for 20 duplicate lneasurenlents tions since these have not been reported before. We have
in each sample. * Based on presence and degree of edema in replotted the data from the tables as Fig. I (normalized
the sampled slides. as percent of the base-line control period average) for
each of the three base-line and three experimental peri-
of diffusion and its subdivisions. We made a single deter- ods. Individual mcasurcmcnts of Vc during the base-line
mination of Q, Qi, and Qpevw. At experimental time period never deviated more than 20 % from the average
zero in g dogs whole blood and dextran infusions were and in 60 of the 81 individual determinations the devia-
begun, occasionally supplclnented by dextran containing tion was less than 10 ‘?. The standard deviation of the
norephinephrine in order to change left ventricular end- sample is 7.5 70.
diastolic pressures the desired amount. These animals are During the experimental periods Vc in the five control
referred to as the high-pressure edema series (P). In I 2 animals ‘did not deviate significantly from the base-line
dogs, at time zero, alloxan dissolved in saline was injected average although one individual determination (dog I C,
rapidly intravenously in doses of 50--I 50 mg/kg (in 5-10 experimental period 2) was markedly different. We pre-
ml saline). These animals are referred to as the alloxan sume this is an error because the Vc value was at the
edema series (A). WC used 5 animals spaced irregularly control level in the third period.
through the total scrics as checks on the over-all stability In the high-pressure series Vc increased an average of
of our procedures. In these animals, at experimental time 42 ‘,Z in experimental period I when edema as measured
zero, saline (equal - in volume to that used in the alloxan by AQpevw was minimal. The increase for the group is
injections) was injected rapidly intravenously. These ani- significant (P < 0.01). When the first experimental
mals arc referred to as the control series (C). period changes in Vc are plotted against the changes in
Following the various injections we waited IO rnin, Plved for the controls and the P series the results arc also
then made three more paired determinations of diffusing significant (rank correlation coefficient --t-0.77, P <
capacity and its subdivisions; one after 10-20 min (exper- 0.01). The P series group average increases further in ex-
imental period I), one after 20.-30 min (experimental perimental periods 2 and 3 but the scatter is very large
period 27, and one after 30-45 min (experimental period and several values are actually decreasing.
3).
. The opening of the thorax following the last diffusion In the alloxan group Vc decreased an average of 16 %
series took approximately 15 min so that most of the in experimental period I when edema as measured by
anatomical specimens were obtained at an experimental AQpevw in 3 animals had not started. The change from
time of about I hr. The studies of Q, Qi, and Qpevw the control period is barely significant (P / 0.05) but is
when made during the experimental period were done highly significantly different from the P series (P < 0.01).
after the first and third Vc determinations. The A series group average decreased further in experi-
mental periods 2 and 3.
RESULTS
The principal functional data are summarized in At thoracotomy, the control series lungs were well
Table I (controls), Table 2 (high-pressure edema), and inflated and orangy-pink in color. The color intensity was
STAUB, NAGANO, AND PEARCE
FIG. 3. High-pressure edema. A: fresh-frozen section from lower Many alveoli are fluid filled. Linear streaks in the free fluid are ice
zone showing moderately severe edema. Two large circular open- crystal spaces due to suboptimal freezing rates. Alveolar fluid con-
ings are terminal bronchioles. Smaller black openings are respira- tains red blood cells. Fluid-filled alveoli are at reduced volume
tory bronchioles and alveolar ducts. Most of the alveoli are filled (alveolar walls folded). Air-filled alveoli show either no free
with sanguinous fluid although some (arrows) air-filled ones fluid or small amounts in the corners. (IO p, fixed section). D:
remain. B: earliest stage of edema. Note marked tilling of the
severe edema. All alveoli at this level filled with blood-tinged fluid.
interstitial compartment (IS). No alveolar fluid. Alveolar walls
appear normal. Some red cells in interstitial fluid (IO p, fixed Alveolar capillaries congested and walls markedly folded (50 p,
section). C; moderate edema. Congested alveolar capillaries. fixed section).
slightly greater in the lower than upper zones. On slicing there appeared to be atelectasisg along the sharp lower
the lung there was moderate blood flow from severed zone margin (costophrenic margin) but the amount
vessels. The lung color was even at any given hydrostatic never exceeded I or 2 % of the lobe volume by our esti-
level. There was no evidence of fluid in the airways. mate. Figure 3A is a fresh-frozen thick section showing a
Figure 2A is a thick fresh-frozen section of control lung. moderately edematous lung.
Microscopically, the alveoli and alveolar ducts in all Microscopically, the earliest evidence of excess fluid in
regions were well inflated and the airways clear. There a region was widening of the perivascular and peribron-
was no evidence of excess fluid in the perivascular and chial connective tissue spaces and distention of the lym-
peribronchial connective tissue spaces (Fig. 2B) and no phatics (IS in Fig. 3B). In those lungs showing mild to
alveolar fluid (Fig. 2C). moderate quantities of edema, the fluid both in the
Table 4 lists measurements of total alveolar wall thick- alveoli and the tissue spaces and lymph vessels was not
ness in the upper and lower zones of two control animals. grossly bloody but always contained some red blood cells
In the high-pressure series the lungs at thoracotomy (Fig. 3C). In the animals with the highest pulmonary
were plethoric. Gross evidence of edema seen from the vascular pressures there was evidence of considerable red
surface was usually confined to the lower 0-30 % of the cell leakage into the lower zones (Fig. 30). The lower
lobe depending on how high the vascular pressure had margin atelectasis described grossly was confirmed in the
been. These lower zones were dark red-purple, smooth, two animals.
and heavy. The upper zones were well inflated. Even in
9 Atelectasis refers to completely empty alveoli. The walls are
the most severely edematous lungs there was little foam folded and closely opposed, so there is no material (gas or liquid)
in the large airways; however, when the lobe was sliced, in the alveolar space. The best example is the completely degassed
pink-tinged foam could be expressed from the smaller lung. The important differentiation is between atelectasis and the
fluid-filled alveolus. The determination of atelectasis is properly a
airways. Much more blood flowed from the severed ves- microscopic one, but gross areas of lung that are irregularly de-
sels than in the control group. In two of the nine animals pressed are usually aletectatic.
ACUTE PULMONARY EDEMA 235
TABLE 5. Dz$erences between elevated hydrostatic ;f,ressure edema and alloxan edema
High Pressure Alloxan
Pulmonary capillary pressure above 20-25 mm Hg Pulmonary capillary pressure below edemogenic level (even if Ppa =
pcq-4
Pressure persistently elevated as edema develops Pressure transiently elevated during first 2-5 min; edema develops 30-60
min later
Pulmonary capillary blood volume increased Pulmonary capillary blood volume decreased
Cardiac output usually increased Cardiac output not much altered after initial transient phase
Central blood volume increased Central blood volume decreased
Edema formation progressive as long as pressure ele- Edema formation often fulrninant
vated
Alveolar capillaries throughout lung congested with Alveolar capillaries congested with red blood cells in lower zone; upper
red blood cells zone normal or anemic
Red cells in edema fluid: gross bleeding in severe No red cells in edema fluid; no bleeding even in severest edema
edema
Fluid-filled alveolar size decreased ; moderately Fluid-filled alveolar size decreased; slightly folded alveolar walls; no
folded walls; occasional marginal atelectasis atelectasis
volume and elevation of Plved, the initiating factor is the rable levels of edema in the two groups as judged by
elevation of capillary hydrostatic pressure (I 7, 41). We Qpevw, A(A-a)DOz, or grossly after experimental period
cannot completely rule out increase in capillary surface 3. Aviado and Schmidt (4) reported a decrease in pul-
area for fluid exchange (VC increased) as a concomitant monary capillary blood volume in the first IO min after
factor but in other conditions, such as exercise, Vc may alloxan. They tagged the circulating red blood cells with
increase two- to three-fold without any evidence of edema 32P and measured radioactivity with a Geiger counter
( 20 ) . over a spot on the surface of the lung. From anatomical
In the edema induced by the intravenous injection of considerations, it is certain that they were measuring
large quantities of alloxan there has been controversy as small artery and vein volume in addition to capillary
to the initiating factors. Aviado and Schmidt (4) orig- volume. Their results, although they appear the same as
inally believed it was persistent pulmonary venous con- ours, bear more resemblance to our combined finding of
striction leading to increased capillary hydrostatic pres- decreased Vc and central blood volume (see also 27).
sure. However, Gruzhit and his coworkers (16) and Part of the decrease in Vc after alloxan could be due
Fejfar and his associates (I I) presented evidence that to uneven distribution of capillary blood volume per unit
increased capillary permeability was the key factor. alveolar gas volume. This occurs normally to some extent
Aviado (3) has modified his view recently to include both (6) and is explained by the interrelations of alveolar
factors although he still favors venular constriction. pressure to pulmonary capillary pressure on a hydro-
Our results support the increase in capillary permea- static basis. It should not cause any significant error if
bility as the chief factor in alloxan edema. Table 5 the alveolar sampling is proportional to alveolar volume
summarizes the major points of difference between the (12) or if th ere arc no changes during an experiment. Our
two forms of edema. The differences include both psysi- histologic observations of decreased red cell content in
ologic and histologic items. the upper zone alveolar capillaries compared to con-
Elevation of vascular volume by transfusion leads to gested capillaries in the basal edematous zone suggests
prolonged increase in the pulmonary capillary pressure increased unevenness of Vc distribution; however, we
as judged by the persistent elevations of Plved and Ppa. have no quantitative evidence on the extent of this effect
Guyton and Lindsey (I 7) found that edema did not and its contribution to the measured decrease in Vc.
develop until pulmonary capillary pressure as measured In the APPENDIX we present some data on the single-
by left atria1 pressure exceeded 24 mm Hg in dogs with breath distribution of regional ventilation and Vc (as
normal plasma oncotic pressure. In our experiments judged by perfusion) in seven dogs before and during the
edema developed only in those animals with a persistent development of alloxan edema. The results indicate an
left ventricular end-diastolic pressure above 20-25 mm increase in relative ventilation to perfusion in the upper
Hg. III the alloxan experiments we found that maximum zone. The immediate decrease in Vc as measured by us
pulmonary capillary pressure (taken as = Ppa) was could be accounted for by this shift. The data compare
nowhere near this value except in the transient state (first favorably with the histologic picture between upper and
2-5 mm). This pattern is the same as Gruzhit (16) and lower zones in these animals.
Aviado (3) found. In summary, regardless of whether the Vc measure-
The rneasurement of Vc shows important differences ment is exact or not in such a changeable situation as
between the two series; Vc increased in the high-pressure acute edema, the fact that we found marked differences
group
c and decreased in the alloxan group. It could be in Vc is new evidence that the mechanisms of edema in
argued that edema might mask Vc increases after alloxan the two series are different.
as indeed it must when fluid filling prevents ventilation The differences in total pulmonary flow and central
of basal alveoli, but this cannot be a major factor in the blood volume are additional evidence that the two groups
first 10-20 min (experimental period I) and it certainly are different in mechanism, although flow changes alone
does not explain the differences in Vc between compa- do not lead to edema and a decrease in central volume
ACUTE PULMONARY EDEMA 237
TABLE 6. Relative ventitation of uj$er and lower zone of lung Even more striking is that in any single microscopic
in seven dogs before and after rabid field each alveolus was either normal (air filled) or com-
intravenous injection of alloxan pletely filled by edema fluid and at a somewhat reduced
I volume, as judged by folding of the alveolar wall. The
Exp Control IO min 20 min 30 40 j 50 60 Anatomy lack of intermediate forms suggests that filling is rapid
X-0. ) Period min min , mm min
-- I ____- -I-- and the intermixing of normal with edema-filled alveoli
B / -139 . ‘67 0’94 . I69 No gross edema suggests that filling is quanta1 in nature, that is, individ-
D i l 704 * 750 .842 ,815; No gross edema ual alveoli fill essentially independently of their neigh-
A / .311 *35o 0389 Moderate bors. This view has previously been suggested by Cook
edema
and co-workers (9) from their observations on the me-
C 2.65 2.40 I *74 2.60 Moderate
edema chanical behavior of edematous lungs. It is also consistent
I: I .468 ,723 .841 Moderate with the findings of Williams (44) on the A(A-a)DOz in
edema edema and Forster’s mention of his unpublished diffusion
.363 0372 Severe edema data in cats (12).
I .oo .g22 Severe edema
Figure 5 shows the complete sequence of events. Fol-
Data expressed as ratio of upper zone to lower zone counts. lowing the filling of the interstitial compartment (Fig.
Ratios along any line are such that an increase indicates rela- 5B) fluid continues to accumulate in the alveolar walls
tively more upper zone ventilation. * Second injection of which, however, have a limited capacity to hold addi-
alloxan.
tional fluid. Fluid leaks into the alveolar spaces at the
corners where curvature is high (Fig. 5C). Bccausc filling
that fluid enters and flows over the alveolar epithelial is from the bottom up the alveolar interfacial surfactant
sirface. layer is lifted upward. As fluid continues to accumulate
Eventually fluid leakage, if it exceeds lymphatic trans- the radius of curvature across the air-liquid interface
port capacity, will fill the interstitial compartment (that decreases and by Laplace’s equation there will be some
is, reduce the hydrostatic gradient). Additional fluid critical interfacial tension and curvature at which the
leakage must then remain at the alveolar level. existing transpulmonary pressure will no longer be able
9 Alveolar wall edema. The data in Table 4 show a to hold the alveolus at a stable configuration (9). The
progressive increase in total alveolar wall thickness as alveolus changes rapidly to a new lower volume with a
edema becomes more severe. The degree of thickening is new curvature and tension at the gas-fluid interface (Fig.
similar for both types of edema so it is unlikely that it 5D). The actual final volu~nc of an alveolus compared to
can be entirely accounted for by expansion or dilation of its normal air volume at a given Ptp depends on the
capi llaries. Schulz (34) has shown by electron microscopy interfacial tension at the critical configuration and the
that the components of the alveolar-capillary membrane rapidity with which fluid in the alveolar walls, capillaries,
progressively thicken. Our data indicate that, at most, and adjacent alveoli flows into the collapsing alveolar
the over-all alveolar wall thickness increase is about 2 p. gas space.
If the entire increase is attributed to the membrane over WC still do not know whether the normal surface
the capillaries rather than to an increase in capillary tension in an alveolus is altered during edema formation.
luminal diameter, then the maximum increase in the Unfortunately, the high surface tension found in pul-
alveolar-capillary membrane is about I p on each side of monary edema fluid is not useful as evidence (32). We
the wall. Since the average alveolar-capillary membrane need to know the actual interfacial tension in the indi-
thickness exclusive of plasrna within the capillary is be- vidual alveoli, not that of the bulk phase. If normal
tween 0.5 p (Weibel and Knight (42), rat lung) and 0.8 alveolar surface tension were o dynes/cm there would
p (Meesen (23), human lung) the increase in severest have to be a significant increase before quanta1 filling
edema would, at most, be threefold. From theoretical occurred. Substances in plasma can inhibit alveolar sur-
calculations of the effect of increasing alveolar-capillary factant (I). On the other hand, if the normal alveolar
membrane thickness on gas exchange we could not expect surface tension is 10-20 dynes/cm (8, 26), quanta1 filling
alveolar wall edema by itself to contribute significantly could occur without significant changes in that tension.
to a change in the arterial blood oxygen sa turation or There is indirect evidence that suggests surface tension
the A .(A-a)DOz alth ough it would cause some decrease in need not be changed. Pulmonary edema foam has very
the measured DL (36).
good surface-tension lowering properties (25). Tierney
3. FiZZing of alveoli. Except for red cells in the edema
and Johnson (40) found that addition of plasma under-
fluid of high-pressure edema the findings in the alveolar
spaces of both types of edema were similar. The absence neath an already formed surface-active layer did not
of trapped gas or bubbles within alveoli is strong evidence significantly interfere with the minimal surface tension
for an orderly process of alveolar filling. This anatomic developed. Interference occurred only after plasma was
finding confirms the physiologic findings of Cook (9) and mixed with the surface film. The deflation pressure-
is in contrast to the view of Sharp et al. (35) who pictured volume curves of edmatous dog lungs are nearly normal
alveoli being “closed” to ventilation by a fluid film at (9) which should not be the case if alveolar surface
the mouth. tension were increased. Finally, Said and co-workers (32)
ACUTE PULMONARY EDEMA 139
TABLE 7. Relative perfusion of uj$er and lower zone of lung Several questions remain to be clarified. What is the
in seven dogs before and fohwing rapid actual leakage site in the pulmonary vessels? What is the
intravenous injection of alloxan pathway from alveolar walls to interstitial space? Is there
~-- - really any free water in the alveolar space normally (22)?
30 40 60 How are fluids deposited in alveoli cleared? What are the
$v
.A 0.
10 min 20 min
min min 50 min min Anatomy
- -1
I actual pressures in the interstitial compartment? Can
B 0 0 0 ‘No gross edema clearance of fluid from the interstitial compartment be
D .271 . I67 No gross edema speeded up acutely thus preventing alveolar filling?
A
C I
*309
.18
-358
.300
l 930
-33’
-7’9 * 393
Moderate
Moderate
edema
edema We acknowledge the extensive assistance of Drs. Yanosuke
F . 112 l 257 Moderate edema Sagawa and Takeshi Nakamura in the preparation and conduct
E 0’94 .265 0375 1Severe edema of the experiments, and thank Mrs. Elizabeth Probert for tech-
G .466 * .818 .528i Severe edema
.412 -370 nical assistance during the experiments and for preparation of all
- - the histological specimens.
Data cxprcssed as ratio of upper zone to lower zone counts.
Ratios along any line are such that an increase indicates rela-
APPENDIX
tively more upper zone perfusion. * Second injection of
alloxan given at 40 min. In an attempt to clarify the cause of the measured decrease in
Vc immediately after alloxan we determined the relative distribu-
showed that a degassed lobe could develop edema even tion of ventilation and perfusion in seven additional dogs. We used
the xenon 133 method of Ball et al. (5). We placed two scintillation
though there was no air-liquid interface at all.
counters horizontally over the right side of morphine-chloralose
Functionally, the only important gas-exchange defect anesthetized dogs lying on their backs. The counters were posi-
in pulmonary edema is the marked shuntlike effect that tioned with the aid of lateral chest radiographs. They were colli-
develops owing to fluid-filled alveoli (41, 44). This read- mated so that the upper counter saw only the anterior portion of
ily explains why forced inflation in the lungs in edema the upper and middle lobes and the lower counter saw only the
posterior portion of the lower lobe. The vertical distance between
can improve the arterial oxygen saturation. Since the counters was 8 cm. The outputs of the scintillation rate meters
fluid-filled alveoli are at a reduced volume, increases in were recorded on a Grass polygraph. Approximately I mc ls3Xe
Ptp would, if sufficient, force air into those alveoli (with- was added to 400 ml air in a large plastic syringe and given as a
out necessarily displacing fluid) and permit oxygenation single rapid inflation followed by a IO- to 20-see breath-holding
period. The animal then was given 2 min for washout, after which
of the blood in capillary beds. The danger in this rnaneu-
approximately I-2 mc xenon 133 dissolved in 2 ml saline was
ver, however, as pointed out by Cook (9) and seen in injected into the superior vena cava while the lungs were inflated
some of our own animals, is that, on deflation, air and with 400 ml plain air. We made two measurements of relative
fluid may mix together, leading to foaming and serious ventilation and perfusion during a 2o-min control period, then
obstruction of all airways. injected alloxan (I oo mg/kg) rapidly intravenously and made
determinations of ventilation and perfusion distribution at IO, 20,
Finally, we offer a hypothesis on the interrelation of and 30 or 40 min. Since we did not determine the absolute volume
atelectasis and edema. If the capillaries are not leaky and of lung seen by our counters we can only compare changes in
surface tension is relatively high, then alveoli collapse to ventilation or perfusion.
the atelectatic state. If the capillaries are very leaky and The results are given in Tables 6 and 7. In the first 20 min after
alloxan there was a small shift in ventilation to the upper zone in
surface tension is relatively low, then the alveoli will fill
six of seven dogs regardless of the extent of edema found later at
with fluid at normal volume. If the capillaries are mod- autopsy. The results for perfusion were variable. Upper-zone
erately leaky and surface tension is intermediate, then perfusion decreased in three and increased in three. One dog never
the alveoli will be fluid filled but at a volume less than showed any detectable upper-zone perfusion. The shifts were not
related to the extent of edema.
in the air-filled state.
These results indicate an increase in the unevenness of distribu-
tion of ventilation to perfusion after alloxan. If regional flow is
Xormal Movement of Fluid in the Lung taken as an index of regional Vc (20), then, since the shift of
ventilation to the upper zone W(LS greater in five of seven dogs than
Although the normal bulk movement of fluid through the shift of perfusion, the single-breath DL method would show a
the pulmonary lymphatics is small, its composition is like decreased Vc if the alveolar sample fraction from the upper zone
that of edema fluid except for a lower protein concen- increased. The initial decrease in Vc seen in Table 3 could be
accounted for on this basis.
tration (I o, 41). There is no reason to think that the flow
Our data on relative perfusion are different from those recently
of fluid from capillaries to lymphatics differs between reported by West, Dollery, and Heard (43). They found in acute
these two conditions. Thus fluid leaking from normal edema due to elevated pulmonary hydrostatic pressure that there
alveolar vessels (capillaries or venules) moves through the was a marked shift of perfusion to the upper zones which was
alveolar walls to the lymphatic capillaries in the inter- rapidly reversible when pulmonary venous pressure was lowered.
Comparison of their results with ours supports our view that the
stitial space. The fluid does not enter the alveolar spaces. primary factors leading to increased fluid leakage in the two types
The driving force for the fluid movement appears to be of edema are different. It should also be noted that our histologic
a rather complex matter, but current indirect evidence studies show similar filling of the interstitial compartment in the
and theoretical discussions (18, I g, 2 I, 28, 37) suggest two edemas. This means that the cause and effect relation they
claimed for interstitial edema and pulmonary vascular resistance is
that there is a subatmospheric pressure in the interstitial
of doubtful validity.
compartment. No actual measurements of this pressure We are currently restudying the problem with improved tech-
have been published. nique.
240 STAUB, NAGANO, AND PEARCE
REFERENCES
I. ARRAMS, M. E., AND F. B. TAYLOR, JR. Isolation and quantita- 24. OGILVIE, C. M., K. E. FORSI’ER, W. S. BLAKEMORE, AND J. W.
tive estimation of pulmonary surface active lipoprotein and its MORTON. ,4 standardized breath holding technique for the
interaction with fibrinogen. ~/~~siaZ~~ist 7 : 78, I 964. clinical measurement of the diffusing capacity of the lung for
2. ALYXCIIULE, M. D. Acute Pulmonary Edema. New York: Grune carbon monoxide. J. CZin. Invest. 36 : I -I 7, I 957.
and Stratton, 1957, p. I I -26. 25. PAI-TLE, R. E. Properties, function and origin of the alveolar
AVIADO, D. M. ‘I’hcr Lung Circulation. New York: Pcrgamon, lining layer. Nature I 75: I 125-1 I 26, 1955.
1965, vol. 2 p. 878.-886. 26 PATTT.F, R E Surfare lining of IIIII~P; alveoli. PhysioZ. Rev. 45:
AVIAI)O, D. M., AND C. F. SCHMIDT. Pathogenesis of pulmonary 48-79, 19%
~&III;\ by alloxan. Circulation Res. 5 : 180-186, 1~57. 17. PEARCE, M. L., . J. Y AMASLIITA, AND . J. BEAZELL. Measureinent
BALL, W. C., P. B. STEWARD, L. G. S. NEWSHAM, AND D. V. of pulmonary edema. Circulation Res. I 6 : 482-488, I 965.
BAIXS. Regional pulmonary function studied with xenon I 33. 28. PERMUU, S. Effect of interstitial pressure of the lung OII pul-
J. CZin. Invest. 41 : 5 I g-53 I, I 962. monary circulation. Med. Thorac. 22 : I I 8-13 I, I 965.
6. BURROWS, K., A. H. NIDEN, C. WI-I-TMAN, R. C. TALLEY, AND 29. RAMSEY, L. I-I., W. PUCHETT, A. JOSE, AND W. W. LACY.
W. R. BARCLAY. Non-uniform pulmonary diffusion as demon- Pericapillary gas and water distribution volumes of the lung
stratccl by the carbon monoxide equilibration technique : calculated from multiple indicator dilution curves. Circulation
experimental results in man. J. CZin. Invest. 3g : 943-51, I 961. Res. 15 : 275-286, 1964.
7. CHINARD, F. P., AND T. ENNS. Transcapillary exchange of 30. ROUGIUON, F. J. W., AND R. E. FORS’IER. Relative importance
water in the dog. Am. J. Physiol. I 7’8 : 197-202, I 954. of diffusion and chemical reaction rates in determining rate of
8. CLEMENTS, J. A. Pulmonary edema and permeability of exchange of gases in the human lung, with special reference to
alveolar membrane. Arch. Environ. Health 2 : 280-283, I 961. true diffusing capacity of pulmonary membrane and volume of
9. COOK, C. D., J. MEAD, G. L. SCHREINER, N. R. FRANK, AND blood in the lung capillaries. J. APPZ. PhysioZ. I I : 290-302,
J. M. CRAIG. Pulmonary mechanics during induced pul- 1957.
monary edema in anesthetized dogs. J. Ap~l. Physiol. I 4: I 771 3’0 RUSZNY~K, I., M. FOLDI, AND G. SZABO. Lymphatics and Lymph
186, =g59* Circulation. New York: Pergamon, 1960, p. 593-633.
I 0. DRINKER, C. K. Pulmonary Edema and In.ammation. Cambridge, 32. SAID, S. I., M. E. AVERY, R. K. DAVIS, C. M. BANERJEE, AND
MLass. : Harvard Univ. Press, I 945. M. EL-GOHARY. Pulmonary surface activity in induced pul-
I I. FEJI:AR, Z., F. ZAJ~C, AND M. FEJFAROV~;. Alloxan induced rnonary edema. J. CZin. Invest. 44 : 458-464, I 965.
experimental pulmonary edema. I. Haemodynarnic altera- 330 SAID, S. I., J. W. LONGACHER, JR., R. K. DAVIS, C. M.
tions. Cor et Vasa I : 56-72, 1959. BANERJEE, W. M. DAVIS, AND W. J. WOODDELL. Pulmonary
12. FORS'I'ER, R. E. Exchange of gases between alveolar air and gas exchange during induction of pulmonary edema in anes-
pulmonary capillary blood : pulmonary diffusing capacity. thetized dogs. J. Aj$Z. Physiol. I g : 403-407, I 964.
Physioz. k?V. 37 : 391-452, 1957. 34. SCHULZ, H. The Submicroscopic Anatomy and Pathology of the
‘3. GOLDBERG, H., L. BENTIVOGLIA, AND X. SAGARMINAGA. Lun,.q. Berlin: Springer, I 959, p. 86-94.
Mechanisms of pulmonary edema. In: Edema: Mechanisms and 35. SHARP, J. T., G. T. GRIFE’I’I‘H, I. L. BUNNELL, AND D. G.
1Managemen t , edited by J. H. Moyer and M. Fuchs. Phila- GREENE. Ventilatory mechanics in pulmonary edema in man.
delphia: Saunders, I 960, p. 708-72 I. J. CZin. Invest. 37 : I I I -I I 7, 1958.
‘4. GRAY, F. D., JR., AND A. S. FIELD, JR. Fluid interchange at the 36. SrAuB, N. C. The alveolar-arterial oxygen tension gradient
pulmonary capillary wall in lung edema. Yale J. RioZ. Med. due to diffusion. J. AppZ. Physiol. 18 : 673--680, 1963.
34: 75-96, 1961. 37. STAUB, N. C. The interdependence of pulmonary structure and
‘5. GREENE, D. G. Pulmonary edema. In: Handbook of PhysioZogy. function. Anesthesiology 24 : 83 I -854, I 963.
Respiration. Washington, D.C. : .&-n. Physiol. Sot., 1964, vol. II, 38. STAUB, N. C., AND W. F. STOREY. Relation between morpho-
sect. 3, chapt. 70, p. 1585-1600. logical and physiological events in lung studied by rapid
16. GRUZHIT, C. C., B. PERAL~I’A, AND G. K. MOE. The pulmonary freezing. J. AppZ. Physiol. I 7 : 38 I -390, I 962.
arterial pressor effect of certain sulfhydryl inhibitors. J. 39. STOREY, W. F., AND N. C. SY’AUB. Ventilation of terminal air
Pharmacol. ExptZ. Therap. I 01 : I 07-1 I I, I g5 I. units. J. APPZ. Physiol. I 7 : 3g I -397, I 962.
‘7. GUYTON, A. C., AND A. W. LINDSEY. Effect of elevated left 40. TIERNEY, D. F., AND R. P. JOHNSON. ,4ltered surface tension of
atria1 pressure and decreased plasma protein concentration on lung extracts and lung mechanics. J. APPZ. PhysioZ. 20: I 253‘-
the development of pulmonary edema. Circulation Res. I : 64g- I 260, 1965.
6579 1959. 4’. VISSCHER, M. B., F. J. HADDY, AND G. SI’EPHENS. The physiol-
18. HAYEK, 13. VON. Die Menschliche Lunge. Berlin : Springer, 1953. ogy and pharmacology of lung edema. Pharnracol. Rev. 8 : 38g-
Translation by V. E. Krahl, New York: Hafner, 1960. 434, ‘956.
‘9. HOWELL, J. B. L., S. PERMUTPI’, D. F. PROCTOR, AND R. L. 42. WEIBEL, E. R., AND B. W. KNIGH’T. A morphometric study on
RILEY. Effect of inflation of the lungs on different parts of the the thickness of the pulmonary air-blood barrier. J. CeZZ BioZ.
pulrnonary vascular bed. J. APPZ. Physiol. I 6 : 7 I -76, I g6 I. 2 I : 367-396, 1964.
20. JOHNSON, R. L., JR., W. S. SPICER, J. M. BISHOP, AND R. E. WEsr, J. B., C. T. DOLLERY, AND B. E. HEARD. Increased
43.
'FORSTER. Pulmonary capillary blood volume, flow and diffus- pulmonary vascular resistance in the dependent zone of the
ing capacity during exercise. J. AppZ. Physiol. 15 : 893-902, isolated dog lung caused by perivascular edema. Circulation
I 960.
Res. 17: 191-206, 1965.
21. MACKLIN, C. C. Transport of air along sheaths of pulmonic
44. WILLIAMS, M. H., JR. Effect of ANTU-induced pulmonary
vessels from alveoli to mediastinum. Arch. Internal Med. 64 : g I 3-
edema on the alveolar-arterial O2 pressure gradient in dog.
92% 1939.
22. MACKLIN, C. C. Pulmonary sumps, dust accumulations, Am. J. Physiol. I 75 : 84-86, 1953.
alveolar fluid and lymph vessels. /I& An~t~mic~ 23 : I 33, I 954. 45. YT)IJNG, R. Cl., JR., H. NAGANC), T, R. VAUGHAN, *JR., AND N.
MEESSEN, H. The pathomorphology of diffusion processes in C. S’rAuB. Pulmonary capillary blood volume in dog: effects of
23.
the lung. Stanford :Wed. BUZZ. I g : I g-3 1, I g6 1. 5hydroxytryptamine. J. AppZ. Physiol. 18 : 264-268, 19613.